作者: Hanane Omichessan , Gianluca Severi , Vittorio Perduca
DOI: 10.1371/JOURNAL.PONE.0221235
关键词:
摘要: Mutational signatures refer to patterns in the occurrence of somatic mutations that might be uniquely ascribed particular mutational process. Tumour mutation catalogues can reveal but are often consistent with spectra produced by a variety mutagens. To date, after analysis tens thousands exomes and genomes from about 40 different cancer types, characterized unique probability profile across 96 trinucleotide-based types have been identified, validated catalogued. At same time, several concurrent methods developed for either quantification contribution catalogued given sequence or identification new sample sequences. A review existing computational tools has recently published guide researchers practitioners through their signature analyses, other introduced since its publication and, systematic evaluation comparison performance such is still lacking. In order fill this gap, we carried out an empirical main packages available using both real simulated data. Among results, our study shows more difficult cancers multiple each having small contribution. This work suggests detection based on probabilistic models, especially EMu bayesNMF, general better than NMF-based methods.